Project Summary: Neural computation requires the coordinated effort of thousands of interrelated and often genetically similar neurons. These neurons form physically intermingled networks and subnetworks that act together to amplify and strengthen sensory perceptions or select motor action. Such co-active ensembles are known to be preferentially interconnected, and may represent a functional element of neural processing with unique properties, such as pattern completion and competition between ensembles. In this proposal I will gain a mechanistic understanding of how ensembles of co-active neurons interact by probing the function of individual and groups of neurons in an awake mouse. I will examine: how ensembles of pyramidal cells interact with other pyramidal cells and local inhibitory neurons in a visual task, how these ensembles influence motor behavior, and how specific ensembles respond to information from other cortical areas. Despite the potential importance of ensembles in cortical coding, the intermixed nature of these groups has made them particularly hard to study. While conventional optogenetic techniques can manipulate genetically identified neurons in a region, they are incapable of selectively manipulating intermingled neurons that differ only by their functional properties. Critically, new multiphoton optogenetic techniques are beginning to allow manipulation of cells chosen by their activity alone, however such techniques require further development. In the K99 phase of this proposal, I will continue my training through the development of novel optical systems for multiphoton stimulation and through use of these technique understand cortical function. By combining these new optical techniques with novel opsins designed for in vivo multiphoton use that I have already developed, I now have the ability to write in or edit neural activity across many neurons with a precision never before possible. By altering ensemble activity during visual perception I will determine the causal contributions of individual neurons as well as populations of neurons to sensory coding. In the R00 phase, through manipulations in motor cortex I will unravel the behavioral impact of these groups, probing the role motor ensembles play in motor action, and study how neurons interact across modalities. The ability to both edit and monitor the activity of neural subnetworks is critical to gaining a mechanistic understanding of perception and action. The conclusions we draw from this proposal will help to describe how all information is presented in the cortex, but can only be reached with advanced techniques.